Simulating Nonlinear Neutrino Oscillations on Next-Generation Many-Core Architectures

TL;DR

This paper introduces XFLAT, a simulation code for neutrino oscillations in supernovae, optimized for next-generation many-core architectures, demonstrating scalable performance on heterogeneous supercomputers.

Contribution

Development of XFLAT, a parallel astrophysical simulation code that efficiently utilizes MPI, OpenMP, and SIMD on CPUs and Xeon Phi co-processors.

Findings

XFLAT achieves scalable performance on multi-node supercomputers.

The code effectively utilizes SIMD vectorization for neutrino oscillation simulations.

I/O and multi-node configurations impact performance on heterogeneous architectures.

Abstract

In this work an astrophysical simulation code, XFLAT, is developed to study neutrino oscillations in supernovae. XFLAT is designed to utilize multiple levels of parallelism through MPI, OpenMP, and SIMD instructions (vectorization). It can run on both the CPU and the Xeon Phi co-processor, the latter of which is based on the Intel Many Integrated Core Architecture (MIC). The performance of XFLAT on configurations and scenarios has been analyzed. In addition, the impact of I/O and the multi-node configuration on the Xeon Phi-equipped heterogeneous supercomputers such as Stampede at the Texas Advanced Computing Center (TACC) was investigated.